Feature request: Add setting to disable tab-to-accept suggestions

I would like to strongly support the request for an option to disable the "Tab to Accept" auto-suggestions.

This feature is currently disruptive to my workflow rather than helpful. Specifically:

• Distraction: The ghost text frequently suggests a thought or direction that I actively disagree with. Seeing incorrect logic appear while I am typing ruins my flow and adds unnecessary cognitive load.

• Disruption: It visually breaks my train of thought and makes the terminal feel unpredictable.

Furthermore, I would argue that this feature should be off by default. It modifies standard terminal behavior significantly enough that it should be an opt-in choice, rather than a default setting that users are forced to disable.

It also takes away the previous tab to toggle thinking shortcut. It removes a potentially useful feature to insert poor-quality slop-grade suggestions. This seems like a poorly designed attempt to automate the use-cases for which slash commands were created.

show me how to shoot myself in the foot (tab to accept)

Probably related and understandable frustration with this feature: #13920

Please add an option to remove this new feature, it's very distracting

It totally breaks the keyboard flow: C-k, C-u, C-c all won't get rid of it. Terrible.

Mentioning #13937

Thanks for the feedback, and to everyone who chimed in here. We hear you on the concerns about this feature being disruptive to workflows.

You can disable prompt suggestions by setting the environment variable: CLAUDE_CODE_ENABLE_PROMPT_SUGGESTION=false

You can add this to your shell profile (.bashrc, .zshrc, etc.) or to your ~/.claude/settings.json under "env":

{
  "env": {
    "CLAUDE_CODE_ENABLE_PROMPT_SUGGESTION": "false"
  }
}

We're taking the feedback here seriously regarding discoverability and whether this should be opt-in vs opt-out. Appreciate you taking the time to share your experiences

@bogini Thank you for the quick turn around on this! We definitely appreciate hearing back from you and the team so fast!

Also if I may voice more of my opinion on "opt-in" vs "opt-out". In general I would say that I prefer "opt-in" way more. It puts the onus on me to discover better workflows and ways of doing things. I will typically blame myself for not spending the time going through configs/reading change logs & docs. It works better because if I find myself doing the same thing over and over or always getting around a problem I can try to solve it on "my time", "when I am ready".

But "opt-in" has the opposite effect, blame shifts more towards "you" and it causes me to have to address the problem "now, regardless if I am ready", just to get back to the state where I was before.

Again thanks for the workaround, its really cool to help shape this awesome product! Keep it up!

Question: does this use my context / tokens to generate the autocomplete suggestions?
If it does, I definitely want to selectively enable it.

Regardless: I would be ready to leave it enabled if I have control over what suggestions it makes. It's currently offering to do things that I never want it to execute.

I have clear rules on certain tasks that must be invoked only by the human in charge of an operation. If it would respect the allow/ask/deny rules, or if I have another field to control it further, then I'd be less alarmed.

In v2.0.71, you can disable prompt suggestions via /config:

1. Type /config in Claude Code
2. Find the "Prompt suggestions" toggle
3. Turn it off

This is in addition to the environment variable method:

// ~/.claude/settings.json
{
  "env": {
    "CLAUDE_CODE_ENABLE_PROMPT_SUGGESTION": "false"
  }
}

This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.